[ 
https://issues.apache.org/jira/browse/CAMEL-27142?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18146581#comment-18146581
 ] 

Guillaume Nodet commented on CAMEL-27142:
-----------------------------------------

This issue is being investigated by a coding agent (on behalf of gnodet).

Initial analysis: OpenAIResponsesProducer currently only records basic-level 
GenAI observability (model, tokens, finish reason). The detailed/full 
observability levels (invoke_agent span around tool loop, request parameters, 
conversation id, cached tokens, messages/instructions/tool definitions) 
introduced by CAMEL-27131 for OpenAIProducer are not yet implemented in 
OpenAIResponsesProducer. A fix session has been dispatched to mirror the same 
pattern.

_Note: This comment was generated by an AI coding agent and requires manual 
verification._

> camel-openai - record the detailed and full GenAI data for the responses 
> operation
> ----------------------------------------------------------------------------------
>
>                 Key: CAMEL-27142
>                 URL: https://issues.apache.org/jira/browse/CAMEL-27142
>             Project: Camel
>          Issue Type: Sub-task
>          Components: camel-openai
>            Reporter: Federico Mariani
>            Assignee: Guillaume Nodet
>            Priority: Major
>
> CAMEL-27131 added the {{detailed}} and {{full}} levels of 
> {{camel.aiObservability.level}}. The {{chat-completion}} operation of 
> {{camel-openai}} records them: an {{invoke_agent}} span around its tool loop, 
> and per model call the request parameters, the conversation id and the cached 
> tokens at {{detailed}}, plus the messages, the system instructions and the 
> tool definitions at {{full}}.
> The {{responses}} operation records at every level what {{basic}} records: 
> one {{chat}} span per model call with the model, the token usage and the 
> finish reason ({{OpenAIResponsesProducer.observedCall}}). Its MCP tool calls 
> get {{execute_tool}} spans from {{McpToolCallExecutor}}, but without an agent 
> span they sit next to the model calls instead of under one agent run.
> Record for the {{responses}} operation what {{chat-completion}} records: the 
> {{invoke_agent}} span around the tool loop, named after the route id, and per 
> model call the request parameters, the conversation id, the cached input 
> tokens and, at {{full}}, the input items as messages, the {{instructions}} as 
> system instructions and the tools.
> _Claude Code on behalf of Croway_



--
This message was sent by Atlassian Jira
(v10.3.26#10030026)

Reply via email to